Comparison of Univariate and Bivariate Parametric Model for Wind Energy Analysis

Abstract

Single parameter or univariate parametric model of wind speed is essential in studying the wind energy potential of an area. But, the joint modelling of wind speed and direction is believed to be much more significant in representing wind regime. In this work, four models of the joint probability distribution of wind speed and wind direction are selected and thoroughly analysed. They are namely Weibull-finite mixture von Mises (fmvM), gamma-fmvM, inverse gamma-fmvM and Burr-fmvM. The proposed bivariate models are constructed by considering the marginal distributions of wind speed and wind direction. The marginal (univariate) case of wind speed modelling based on the conventional distribution. Four of them are selected based on the goodness-of-fit. However, finite mixture Von Mises is selected based on wind direction modelling, as it best for describing the multimodal condition of wind direction in Malaysia. This study reveals that the bivariate parametric model gives slightly higher mean wind power density (W/m2) when compared to univariate model. Thus, the result verified that bivariate parametric model is significant in representing wind regime of an area.